cifkao commited on
Commit
a431ec5
·
1 Parent(s): cb5325e

Added metadata columns, v1.4.0

Browse files
README.md CHANGED
@@ -104,14 +104,14 @@ The evaluation is implemented in our [`alt-eval` package](https://github.com/aud
104
  from datasets import load_dataset
105
  from alt_eval import compute_metrics
106
 
107
- dataset = load_dataset("jamendolyrics/jam-alt", revision="v1.3.0", split="test")
108
  # transcriptions: list[str]
109
  compute_metrics(dataset["text"], transcriptions, languages=dataset["language"])
110
  ```
111
 
112
  For example, the following code can be used to evaluate Whisper:
113
  ```python
114
- dataset = load_dataset("jamendolyrics/jam-alt", revision="v1.3.0", split="test")
115
  dataset = dataset.cast_column("audio", datasets.Audio(decode=False)) # Get the raw audio file, let Whisper decode it
116
 
117
  model = whisper.load_model("tiny")
@@ -123,7 +123,7 @@ compute_metrics(dataset["text"], transcriptions, languages=dataset["language"])
123
  ```
124
  Alternatively, if you already have transcriptions, you might prefer to skip loading the `audio` column:
125
  ```python
126
- dataset = load_dataset("jamendolyrics/jam-alt", revision="v1.3.0", split="test").remove_columns("audio")
127
  ```
128
 
129
  ## Citation
 
104
  from datasets import load_dataset
105
  from alt_eval import compute_metrics
106
 
107
+ dataset = load_dataset("jamendolyrics/jam-alt", revision="v1.4.0", split="test")
108
  # transcriptions: list[str]
109
  compute_metrics(dataset["text"], transcriptions, languages=dataset["language"])
110
  ```
111
 
112
  For example, the following code can be used to evaluate Whisper:
113
  ```python
114
+ dataset = load_dataset("jamendolyrics/jam-alt", revision="v1.4.0", split="test")
115
  dataset = dataset.cast_column("audio", datasets.Audio(decode=False)) # Get the raw audio file, let Whisper decode it
116
 
117
  model = whisper.load_model("tiny")
 
123
  ```
124
  Alternatively, if you already have transcriptions, you might prefer to skip loading the `audio` column:
125
  ```python
126
+ dataset = load_dataset("jamendolyrics/jam-alt", revision="v1.4.0", split="test").remove_columns("audio")
127
  ```
128
 
129
  ## Citation
loader.py CHANGED
@@ -8,7 +8,7 @@ from typing import Optional
8
 
9
  import datasets
10
 
11
- _VERSION = "1.3.0"
12
 
13
 
14
  _CITATION = """\
@@ -83,6 +83,9 @@ class JamAltDataset(datasets.GeneratorBasedBuilder):
83
  }
84
  ],
85
  "language": datasets.Value("string"),
 
 
 
86
  "license_type": datasets.Value("string"),
87
  }
88
  if self.config.with_audio:
@@ -153,6 +156,9 @@ class JamAltDataset(datasets.GeneratorBasedBuilder):
153
  "text": text,
154
  "lines": lines,
155
  "language": _LANGUAGE_NAME_TO_CODE[meta["Language"]],
 
 
 
156
  "license_type": meta["LicenseType"],
157
  }
158
  if audio_path is not None:
 
8
 
9
  import datasets
10
 
11
+ _VERSION = "1.4.0"
12
 
13
 
14
  _CITATION = """\
 
83
  }
84
  ],
85
  "language": datasets.Value("string"),
86
+ "artist": datasets.Value("string"),
87
+ "title": datasets.Value("string"),
88
+ "genre": datasets.Value("string"),
89
  "license_type": datasets.Value("string"),
90
  }
91
  if self.config.with_audio:
 
156
  "text": text,
157
  "lines": lines,
158
  "language": _LANGUAGE_NAME_TO_CODE[meta["Language"]],
159
+ "artist": meta["Artist"],
160
+ "title": meta["Title"],
161
+ "genre": meta["Genre"],
162
  "license_type": meta["LicenseType"],
163
  }
164
  if audio_path is not None:
metadata.jsonl CHANGED
The diff for this file is too large to render. See raw diff
 
subsets/de/metadata.jsonl CHANGED
The diff for this file is too large to render. See raw diff
 
subsets/en/metadata.jsonl CHANGED
The diff for this file is too large to render. See raw diff
 
subsets/es/metadata.jsonl CHANGED
The diff for this file is too large to render. See raw diff
 
subsets/fr/metadata.jsonl CHANGED
The diff for this file is too large to render. See raw diff